build based on 0640473

previews/PR2318/destructure/index.html

@@ -69,7 +69,7 @@
 L2 (generic function with 1 method)
 
 julia> L2(m2) isa Float32
-true</code></pre></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/ea9d95f167b277059ebff3bcf7d2656d2ac8a7c5/src/utils.jl#L566-L608">source</a></section></article><h3 id="Save-and-Load"><a class="docs-heading-anchor" href="#Save-and-Load">Save and Load</a><a id="Save-and-Load-1"></a><a class="docs-heading-anchor-permalink" href="#Save-and-Load" title="Permalink"></a></h3><article class="docstring"><header><a class="docstring-binding" id="Flux.state" href="#Flux.state"><code>Flux.state</code></a> — <span class="docstring-category">Function</span></header><section><div><pre><code class="language-julia hljs">state(x)</code></pre><p>Return an object with the same nested structure as <code>x</code> according to <code>Functors.children</code>,  but made only of basic containers (e.g. named tuples, tuples, arrays, and dictionaries).</p><p>Besides trainable and non-trainable arrays, the state will contain leaf nodes that are not arrays, such as numbers, symbols, strings, and nothing values. The leaf types that end up in the state could increase in the future.</p><p>This method is particularly useful for saving and loading models,  since the state contain only simple data types that can be easily serialized.</p><p>The state can be passed to <a href="#Flux.loadmodel!"><code>loadmodel!</code></a> to restore the model.</p><p><strong>Examples</strong></p><p><strong>Copy the state into another model</strong></p><pre><code class="language-julia-repl hljs">julia&gt; m1 = Chain(Dense(1, 2, tanh; init=ones), Dense(2, 1; init=ones));
+true</code></pre></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/06404731e4c19b66ab6c829cf5521ff5a4e6123f/src/utils.jl#L566-L608">source</a></section></article><h3 id="Save-and-Load"><a class="docs-heading-anchor" href="#Save-and-Load">Save and Load</a><a id="Save-and-Load-1"></a><a class="docs-heading-anchor-permalink" href="#Save-and-Load" title="Permalink"></a></h3><article class="docstring"><header><a class="docstring-binding" id="Flux.state" href="#Flux.state"><code>Flux.state</code></a> — <span class="docstring-category">Function</span></header><section><div><pre><code class="language-julia hljs">state(x)</code></pre><p>Return an object with the same nested structure as <code>x</code> according to <code>Functors.children</code>,  but made only of basic containers (e.g. named tuples, tuples, arrays, and dictionaries).</p><p>Besides trainable and non-trainable arrays, the state will contain leaf nodes that are not arrays, such as numbers, symbols, strings, and nothing values. The leaf types that end up in the state could increase in the future.</p><p>This method is particularly useful for saving and loading models,  since the state contain only simple data types that can be easily serialized.</p><p>The state can be passed to <a href="#Flux.loadmodel!"><code>loadmodel!</code></a> to restore the model.</p><p><strong>Examples</strong></p><p><strong>Copy the state into another model</strong></p><pre><code class="language-julia-repl hljs">julia&gt; m1 = Chain(Dense(1, 2, tanh; init=ones), Dense(2, 1; init=ones));
 
 julia&gt; s = Flux.state(m1)
 (layers = ((weight = [1.0; 1.0;;], bias = [0.0, 0.0], σ = ()), (weight = [1.0 1.0], bias = [0.0], σ = ())),)
@@ -95,7 +95,7 @@
 
 julia&gt; JLD2.jldsave(&quot;checkpoint.jld2&quot;, model_state = s)
 
-julia&gt; Flux.loadmodel!(m2, JLD2.load(&quot;checkpoint.jld2&quot;, &quot;model_state&quot;))</code></pre></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/ea9d95f167b277059ebff3bcf7d2656d2ac8a7c5/src/loading.jl#L112-L172">source</a></section></article><article class="docstring"><header><a class="docstring-binding" id="Flux.loadmodel!" href="#Flux.loadmodel!"><code>Flux.loadmodel!</code></a> — <span class="docstring-category">Function</span></header><section><div><pre><code class="language-julia hljs">loadmodel!(dst, src)</code></pre><p>Copy all the parameters (trainable and non-trainable) from <code>src</code> into <code>dst</code>.</p><p>Recursively walks <code>dst</code> and <code>src</code> together using <a href="../models/functors/#Functors.children"><code>Functors.children</code></a>, and calling <code>copyto!</code> on parameter arrays or throwing an error when there is a mismatch. Non-array elements (such as activation functions) are not copied and need not match. Zero bias vectors and <code>bias=false</code> are considered equivalent (see extended help for more details).</p><p>See also <a href="#Flux.state"><code>Flux.state</code></a>.</p><p><strong>Examples</strong></p><pre><code class="language-julia hljs">julia&gt; dst = Chain(Dense(Flux.ones32(2, 5), Flux.ones32(2), tanh), Dense(2 =&gt; 1; bias = [1f0]))
+julia&gt; Flux.loadmodel!(m2, JLD2.load(&quot;checkpoint.jld2&quot;, &quot;model_state&quot;))</code></pre></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/06404731e4c19b66ab6c829cf5521ff5a4e6123f/src/loading.jl#L112-L172">source</a></section></article><article class="docstring"><header><a class="docstring-binding" id="Flux.loadmodel!" href="#Flux.loadmodel!"><code>Flux.loadmodel!</code></a> — <span class="docstring-category">Function</span></header><section><div><pre><code class="language-julia hljs">loadmodel!(dst, src)</code></pre><p>Copy all the parameters (trainable and non-trainable) from <code>src</code> into <code>dst</code>.</p><p>Recursively walks <code>dst</code> and <code>src</code> together using <a href="../models/functors/#Functors.children"><code>Functors.children</code></a>, and calling <code>copyto!</code> on parameter arrays or throwing an error when there is a mismatch. Non-array elements (such as activation functions) are not copied and need not match. Zero bias vectors and <code>bias=false</code> are considered equivalent (see extended help for more details).</p><p>See also <a href="#Flux.state"><code>Flux.state</code></a>.</p><p><strong>Examples</strong></p><pre><code class="language-julia hljs">julia&gt; dst = Chain(Dense(Flux.ones32(2, 5), Flux.ones32(2), tanh), Dense(2 =&gt; 1; bias = [1f0]))
 Chain(
   Dense(5 =&gt; 2, tanh),                  # 12 parameters
   Dense(2 =&gt; 1),                        # 3 parameters
@@ -112,4 +112,4 @@
 false
 
 julia&gt; iszero(dst[2].bias)
-true</code></pre><p><strong>Extended help</strong></p><p>Throws an error when:</p><ul><li><code>dst</code> and <code>src</code> do not share the same fields (at any level)</li><li>the sizes of leaf nodes are mismatched between <code>dst</code> and <code>src</code></li><li>copying non-array values to/from an array parameter (except inactive parameters described below)</li><li><code>dst</code> is a &quot;tied&quot; parameter (i.e. refers to another parameter) and loaded into multiple times with mismatched source values</li></ul><p>Inactive parameters can be encoded by using the boolean value <code>false</code> instead of an array. If <code>dst == false</code> and <code>src</code> is an all-zero array, no error will be raised (and no values copied); however, attempting to copy a non-zero array to an inactive parameter will throw an error. Likewise, copying a <code>src</code> value of <code>false</code> to any <code>dst</code> array is valid, but copying a <code>src</code> value of <code>true</code> will error.</p></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/ea9d95f167b277059ebff3bcf7d2656d2ac8a7c5/src/loading.jl#L39-L89">source</a></section></article></article><nav class="docs-footer"><a class="docs-footer-prevpage" href="../outputsize/">« Shape Inference</a><a class="docs-footer-nextpage" href="../training/callbacks/">Callback Helpers »</a><div class="flexbox-break"></div><p class="footer-message">Powered by <a href="https://github.com/JuliaDocs/Documenter.jl">Documenter.jl</a> and the <a href="https://julialang.org/">Julia Programming Language</a>.</p></nav></div><div class="modal" id="documenter-settings"><div class="modal-background"></div><div class="modal-card"><header class="modal-card-head"><p class="modal-card-title">Settings</p><button class="delete"></button></header><section class="modal-card-body"><p><label class="label">Theme</label><div class="select"><select id="documenter-themepicker"><option value="documenter-light">documenter-light</option><option value="documenter-dark">documenter-dark</option></select></div></p><hr/><p>This document was generated with <a href="https://github.com/JuliaDocs/Documenter.jl">Documenter.jl</a> version 0.27.25 on <span class="colophon-date" title="Friday 25 August 2023 12:26">Friday 25 August 2023</span>. Using Julia version 1.9.3.</p></section><footer class="modal-card-foot"></footer></div></div></div></body></html>
+true</code></pre><p><strong>Extended help</strong></p><p>Throws an error when:</p><ul><li><code>dst</code> and <code>src</code> do not share the same fields (at any level)</li><li>the sizes of leaf nodes are mismatched between <code>dst</code> and <code>src</code></li><li>copying non-array values to/from an array parameter (except inactive parameters described below)</li><li><code>dst</code> is a &quot;tied&quot; parameter (i.e. refers to another parameter) and loaded into multiple times with mismatched source values</li></ul><p>Inactive parameters can be encoded by using the boolean value <code>false</code> instead of an array. If <code>dst == false</code> and <code>src</code> is an all-zero array, no error will be raised (and no values copied); however, attempting to copy a non-zero array to an inactive parameter will throw an error. Likewise, copying a <code>src</code> value of <code>false</code> to any <code>dst</code> array is valid, but copying a <code>src</code> value of <code>true</code> will error.</p></div><a class="docs-sourcelink" target="_blank" href="https://github.com/FluxML/Flux.jl/blob/06404731e4c19b66ab6c829cf5521ff5a4e6123f/src/loading.jl#L39-L89">source</a></section></article></article><nav class="docs-footer"><a class="docs-footer-prevpage" href="../outputsize/">« Shape Inference</a><a class="docs-footer-nextpage" href="../training/callbacks/">Callback Helpers »</a><div class="flexbox-break"></div><p class="footer-message">Powered by <a href="https://github.com/JuliaDocs/Documenter.jl">Documenter.jl</a> and the <a href="https://julialang.org/">Julia Programming Language</a>.</p></nav></div><div class="modal" id="documenter-settings"><div class="modal-background"></div><div class="modal-card"><header class="modal-card-head"><p class="modal-card-title">Settings</p><button class="delete"></button></header><section class="modal-card-body"><p><label class="label">Theme</label><div class="select"><select id="documenter-themepicker"><option value="documenter-light">documenter-light</option><option value="documenter-dark">documenter-dark</option></select></div></p><hr/><p>This document was generated with <a href="https://github.com/JuliaDocs/Documenter.jl">Documenter.jl</a> version 0.27.25 on <span class="colophon-date" title="Friday 25 August 2023 12:26">Friday 25 August 2023</span>. Using Julia version 1.9.3.</p></section><footer class="modal-card-foot"></footer></div></div></div></body></html>